Two important imaging modalities in nuclear medicine are single photon emission computed tomography (SPECT) and positron emission tomography (PET) in which a fraction of the photons emitted directly or indirectly (through positron annihilation) by a radionuclide distribution within a patient are detected. Typical nuclear medicine studies include but are not limited to whole body, heart, brain, thyroid, gastro-intestinal, and breast (scintimammography and positron emission mammography or PEM) imaging.
Image data acquisition in nuclear medicine presents several challenges in addition to constraints imposed by finite acquisition times and patient exposure restrictions. Most photon energies that are of interest in nuclear medicine are higher than the typical photon energies employed in diagnostic x-ray radiography. In particular, PET involves the detection of pairs of very high energy photons (511 keV) due to annihilation events.
The direction vectors and energies of non-scattered photons that escape the body are assumed to be well-defined. Unfortunately, the emission of photons from the radionuclide source distribution is non-directional and the radiation source distribution itself is typically not well-defined. Scattered photons that escape the body may have their energies and/or direction vectors altered. It is desirable for many applications to discriminate against scatter radiation reaching the detector based on energy and/or direction. A Compton-scattered photon suffers an energy loss and change in direction vector whereas a coherent or Rayleigh scattered photon only has its direction vector altered. It may be desirable to only detect radiation with a limited range of direction vectors.
Imaging systems typically offer poor directional discrimination capability and have finite response times within which to detect events (thereby limiting detection rates). Thus detection systems used in nuclear medicine such as gamma (SPECT) cameras (and sometimes PET cameras) employ focused or unfocused collimators, to help define the direction vectors of detected photons. Compton gamma cameras (based on the detection of one or more Compton scattered photons) and most PET cameras rely on electronic collimation. (The Compton camera design uses one or more relatively thin, planar semiconductor (often Si or Ge) arrays as Compton scatterers. One implementation uses a scintillator gamma camera to detect the Compton-scattered photons. Compton gamma cameras are still being refined.)
The detection format for conventional SPECT, PET, and Compton gamma cameras is “face-on” wherein the radiation entrance surface and readout surface are parallel. The majority of clinical SPECT and PET cameras employ scintillator rather than semiconductor detectors. Although semiconductor detectors may offer superior spatial and energy resolution, scintillators are typically less expensive to grow and process, they are highly reliable, they may offer superior stopping power, and they may offer faster response times (desirable for PET and time-of-flight (TOF) PET). Scintillators are employed based on conversion efficiency (that may be energy-dependent or non-proportional), emission spectrum, decay time and after glow, index of refraction (IOR), density, material-dependent photon cross sections, the presence of natural or induced radioactivity, and manufacturing cost.
Two common detector geometries used in nuclear medicine imaging are the planar detector (SPECT and PET) and the ring detector (PET). A basic gamma camera design employs a large, planar array of scintillation crystals or a single, large, planar scintillation crystal optically coupled to an array of photomultiplier tubes (PMTs). A conventional focused or unfocused collimator is typically mounted to the face of the gamma camera. This imaging system is then positioned such that the region of interest containing the source distribution is within the field of view. It provides a limited degree of spatial resolution and energy resolution while removing some fraction of scattered radiation that would otherwise degrade image quality. Unfortunately a substantial fraction of useful unscattered radiation is also attenuated. (An infrequently used design replaces the conventional collimator with a coded aperture such as a uniformly redundant array aperture that is also based on photon attenuation.) Clinical SPECT systems may use one, two, or three gamma camera detector units.
An alternative (face-on) gamma camera design eliminates the use of scintillator crystals and PMTs with a planar, modular 2-D CdZnTe semiconductor detector manufactured by butting small, 2-D (pixellated) CdZnTe arrays. Drawbacks to employing 2-D CdZnTe arrays capable of high detection efficiency include the difficulty of growing thick CdZnTe crystals with acceptable levels of defects and creating low noise, 2-D array readout structures on CdZnTe crystals.
A limitation of the face-on detection format for SPECT and PET imaging is that properties such as the detection efficiency, spatial resolution, and energy resolution exhibit a noticeable energy dependence. Basic edge-on semiconductor and scintillator detector array designs are being used as alternatives to face-on detectors for x-ray and gamma ray radiography (digital mammography and high energy industrial imaging) and gamma ray imaging in nuclear medicine (PET). Basic edge-on array detector designs are suitable for SPECT and Compton scatter imaging as well as PET imaging.
Cost-effective implementations of edge-on detector modules are needed for clinical nuclear medicine imaging systems. Factors to consider include the material properties and costs, the active detector area and volume, the desired spatial, energy, and temporal resolution, and the readout requirements. Consider an edge-on detector module comprised of one or more basic edge-on semiconductor or scintillator planar detectors (for example, a linear array of scintillator rods coupled to a photodiode strip array). The spatial resolution of a basic edge-on planar detector along the dimension of the aperture is defined by the thickness of the edge unless a collimator is used to restrict the incident radiation along that dimension. Increased spatial resolution requires the use of thinner planar detectors. The number of detector planes (and readout elements) doubles each time the aperture height is halved (the aperture resolution doubles). This forces an increase in the packing density of electronics that resides near the array of basic edge-on detectors. As the number of basic edge-on detectors per detector module increases so does the inactive volume (dead space) due to the thickness of the photodetector readout (for edge-on scintillator detectors) and any gaps between the basic edge-on detectors. The problem of dead space between basic edge-on detector planes is more severe for basic edge-on scintillator array detectors than for basic edge-on semiconductor array detectors. For example, deploying a basic edge-on scintillator detector design for a high resolution PET detector requires a very large number of very thin photodetectors such as Geiger-mode silicon photomultiplier (SiPM) arrays, internal discrete amplification photodetector arrays, avalanche photodiode (APD) linear arrays or position-sensitive APDs (PSAPDs) optically coupled to 1-D arrays of LSO scintillator rods with a 1 mm aperture height; see, e.g., Levin (2004) Nuc. Instr. Meth. Phys. Res. A 527:35-40; Levin (2004) IEEE Trans. Nucl. Sci. Vol. 51, No. 3, pp. 805-810, June 2004.
Although these readout detectors are expected to have a thickness less than 0.5 mm this thickness (dead space) is non-negligible compared to the 1 mm aperture height (and it is still significant even for a 3 mm aperture height) of the LSO scintillator array. This dead space degrades the spatial resolution in one dimension as well as the detection efficiency. The impact of this dead space could be mitigated if the scintillator rod aperture height was much larger. (A significant increase in aperture height would ease the requirements on the thickness (and cost) of the readout detector. The readout detector thickness would only need to be sufficiently thin so that the impact of dead spaces or “gaps” between basic edge-on scintillator detectors or edge-on scintillator detector modules on spatial resolution and detection efficiency is acceptable for the imaging task. Commercial, face-on modular gamma cameras have gaps between the butted detector modules.) Another application that benefits from a thin photodetector readout detector is a face-on, wearable PET detector for small animal brain imaging that uses wafer-thin APD arrays, Vaska P, et al., IEEE Trans. Nucl. Sci. Vol. 51, No. 5, pp. 2718-2722, October 2004.
The number of basic edge-on scintillator or semiconductor detector planes required to assemble an edge-on detector module can be reduced by implementing the techniques developed for measuring the “depth of interaction” (DOI) within face-on scintillator and semiconductor detectors. The benefits of this approach can be illustrated by considering a scenario in which radiation is incident face-on upon the anode or cathode side of a planar semiconductor detector of known depth or thickness (height). The DOI spatial resolution can be determined by measuring either the transit times of electrons and holes to anodes and cathodes, respectively, or the ratio of anode and cathode signals. The semiconductor detector DOI accuracy is affected by parameters such as the detector depth, electron and hole mobility, signal diffusion, and the number of defects (such as traps) in the bulk semiconductor material. (The specific parameters that affect scintillator detector DOI accuracy vary with the DOI measurement technique.) Orient the planar semiconductor detector edge-on to the source of incident radiation. The planar semiconductor detector thickness now defines the maximum height of the edge-on semiconductor detector entrance aperture. The electronically-measured face-on detector DOI positional information now defines the edge-on detector sub-aperture resolution (SAR). The interaction position along the height of the edge-on detector aperture is referred to as the interaction height.